Problem

The throughput fields in BenchmarkSummary and MechanismSummary are
named average_throughput_mbps and peak_throughput_mbps, which
conventionally means "megabits per second." However, the actual
calculation divides bytes_per_second by 1,000,000, producing
megabytes per second (MB/s).

The doc comments also say "megabits per second." The console output
correctly displays "MB/s" but the struct field names, doc comments,
and serialized JSON output are misleading.

There is a comment in the code acknowledging the mismatch:

// Note: The summary field is named _mbps, but the calculation
// in update_summary produces MB/s (Megabytes per second).
// The label reflects the calculation.

If the values were truly megabits per second, the calculation would
need to be bytes_per_second * 8 / 1_000_000. The current formula
produces megabytes per second.

Impact

• Anyone consuming the JSON output and trusting the field name
  average_throughput_mbps would interpret the values as 8x smaller
  than they actually are (1 byte = 8 bits).
• Doc comments are factually incorrect.

Fix

Rename the fields to match the actual unit:

• average_throughput_mbps → average_throughput_mb_s
• peak_throughput_mbps → peak_throughput_mb_s
• Update doc comments from "megabits" to "megabytes per second"
• Remove the workaround comments acknowledging the mismatch

Breaking Change

This renames fields in the serialized JSON output. Any downstream
tool parsing average_throughput_mbps from the JSON will need to
update to average_throughput_mb_s.

Files

• src/results.rs — struct definitions, calculation, display, tests
• src/results_blocking.rs — blocking equivalents
• README.md — JSON example output